[Paraview] running paraview in parallel with a batch script/MPI

Thu Feb 14 20:20:02 EST 2013

Fantastic, I'm sure Alan has told me that in the past but I had surely forgotten!  Thanks!
-- Rich

On Feb 14, 2013, at 5:08 PM, Pat Marion wrote:

In case you don't know about the feature, you can enable "trace" and paraview will record your actions in the GUI and generate python code for you.  It can be a nice way to learn.  You can find lots of information about python scripting and the trace feature on the wiki page, or just go to the Tools menu and enable Trace, then do some actions like opening data and applying filters, then stop trace and you will see the generated python script.

Pat

On Fri, Feb 15, 2013 at 5:30 AM, Cook, Rich <cook47 at llnl.gov<mailto:cook47 at llnl.gov>> wrote:
Looks like I need to study up on PV scripting -- I'm so used to VisIt and EnSight, I was hoping the syntax would be pretty intuitive but some of what you are doing there is not clear to me, just because I'm such a newbie.  Thanks for the clarification. I think this sets my mind at ease and helps my user.

-- Rich

On Feb 13, 2013, at 9:18 PM, Pat Marion wrote:

Right, the decomposition is up to the reader, so you need a reader that is "parallel aware".  If you're starting with a .pvtu file, then you should be all set.  If you look inside the pvtu file, it should just be an xml format that lists all the pieces, and the pieces are .vtu.  Each rank will read a different subset of the .vtu files.  Since pvbatch functions just like pvserver, you  can try your work interactively with pvserver first.  In fact, you can connect ParaView to a parallel pvserver, then paste the python script line for line into ParaView's python console and you should get the same results.

You can color by process ID scalars using pvbatch.  I have script that I use for testing which does exactly that.  Notice that the script creates a Sphere(), you could replace that with your reader.

# testParallelRender.py

print "importing paraview.simple..."
from paraview.simple import *

n_procs = servermanager.vtkProcessModule.GetProcessModule().GetNumberOfLocalPartitions()
print "number of ranks: ", n_procs

print "create pipeline..."
Sphere(ThetaResolution=100, PhiResolution=100)
ProcessIdScalars()
UpdatePipeline()

print "creating view..."
CreateRenderView()

print "create reps..."
Show()
lt = CreateLookupTable(RGBPoints=[0.0, 0, 0, 1, n_procs-1, 1, 0, 0], ColorSpace = "HSV")
SetDisplayProperties(ColorAttributeType=0, ColorArrayName="ProcessId", LookupTable=lt)

print "render..."
Render()

GetActiveView().UseOffscreenRenderingForScreenshots = 0

print "writing..."
WriteImage("coloredSphere.png")

print "Done."

Pat

On Thu, Feb 14, 2013 at 3:16 AM, Cook, Rich <cook47 at llnl.gov<mailto:cook47 at llnl.gov>> wrote:
OK, so what I'm hearing is that the pvbatch process will use its MPI rank to decompose the data correctly.
Does this decomposition work correctly for all data formats?  I suspect not, so is there a way to confirm that decomposition is working for a particular type of data?  In the GUI, I can paint the data with the process ID scalars filter.  Any similar trick for pvbatch?
Thanks!
-- Rich

On Feb 12, 2013, at 11:31 PM, Pat Marion wrote:

Hi Rich,

The command line will be:

mpirun -np 1 /path/to/pvbatch /path/to/script.py

The pvbatch executable is similar to pvserver, except proc 0 doesn't wait for a client connection, it reads the python script specified on the command line and executes the script as if it were instructions from a client.  The script will be read and interpreted on proc 0, but the processing will be carried out by all the satellite procs, just like pvserver.

I'd recommend starting with a script that is very simple, like:

# testSphere.py

from paraview.simple import *

Sphere()
writer = XMLPPolyDataWriter(FileName='sphere.pvtp')
writer.UpdatePipeline()

Try getting that to work using -np 1, then try -np 2, you should find pieces of there sphere written by both processes.

Pat

p.s. If your data is unstructured grid, then your writer would be a XMLPUnstructuredGridWriter.  Also, I'm not sure if the name should be "XMLP..." or "XMLPartitioned..."

On Wed, Feb 13, 2013 at 11:48 AM, Cook, Rich <cook47 at llnl.gov<mailto:cook47 at llnl.gov>> wrote:
Hello, ParaView genii,

I would like to run ParaView to view some data in parallel using the following script from a user as a basis.  I'm not sure how to do this right.  I've never scripted Paraview before, and am thus clueless how to make it work.  Can someone on this list tell me how Paraview scripting with MPI works or point me to a tutorial on your massive tutorial pages?  The data is very large and it would be great to decompose the data across the cluster.

Thanks!

try: paraview.simple
except: from paraview.simple import *
paraview.simple._DisableFirstRenderCameraReset()

filein = XMLPartitionedUnstructuredGridReader( FileName=['/p/lscratchd/bodart1/test_para/duct.020000.pvtu'] )

Slice1 = Slice( SliceType="Plane" )

Slice1.SliceOffsetValues = [0.0]
Slice1.SliceType.Origin = [0.0,0.0,0.0]
Slice1.SliceType.Normal = [0.0,0.0,1.0]
Slice1.SliceType = "Plane"

CleantoGrid1 = CleantoGrid()

CellDatatoPointData1 = CellDatatoPointData()

w.FileName="test_slice.vtu"
w.UpdatePipeline()

--
✐Richard Cook
✇ Lawrence Livermore National Laboratory
Bldg-453 Rm-4024, Mail Stop L-557
7000 East Avenue,  Livermore, CA, 94550, USA
☎ (office) (925) 423-9605<tel:%28925%29%20423-9605>
☎ (fax) (925) 423-6961<tel:%28925%29%20423-6961>
---
Information Management & Graphics Grp., Services & Development Div., Integrated Computing & Communications Dept.
(opinions expressed herein are mine and not those of LLNL)

_______________________________________________
Powered by www.kitware.com<http://www.kitware.com/>

Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView

Follow this link to subscribe/unsubscribe:
http://www.paraview.org/mailman/listinfo/paraview

--
✐Richard Cook
✇ Lawrence Livermore National Laboratory
Bldg-453 Rm-4024, Mail Stop L-557
7000 East Avenue,  Livermore, CA, 94550, USA
☎ (office) (925) 423-9605<tel:%28925%29%20423-9605>
☎ (fax) (925) 423-6961<tel:%28925%29%20423-6961>
---
Information Management & Graphics Grp., Services & Development Div., Integrated Computing & Communications Dept.
(opinions expressed herein are mine and not those of LLNL)

--
✐Richard Cook
✇ Lawrence Livermore National Laboratory
Bldg-453 Rm-4024, Mail Stop L-557
7000 East Avenue,  Livermore, CA, 94550, USA
☎ (office) (925) 423-9605<tel:%28925%29%20423-9605>
☎ (fax) (925) 423-6961<tel:%28925%29%20423-6961>
---
Information Management & Graphics Grp., Services & Development Div., Integrated Computing & Communications Dept.
(opinions expressed herein are mine and not those of LLNL)

--
✐Richard Cook
✇ Lawrence Livermore National Laboratory
Bldg-453 Rm-4024, Mail Stop L-557
7000 East Avenue,  Livermore, CA, 94550, USA
☎ (office) (925) 423-9605
☎ (fax) (925) 423-6961
---
Information Management & Graphics Grp., Services & Development Div., Integrated Computing & Communications Dept.
(opinions expressed herein are mine and not those of LLNL)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20130215/af4d24eb/attachment-0001.htm>